Neurology Publish Ahead of PrintDOI: 10.1212/WNL.0000000000012440
Diabetes Mellitus and Cognition: A Pathway Analysis in the MEMENTO Cohort
Author(s):
Eric Frison, MD PhD1,2; Cecile Proust-Lima, PhD3; Jean-Francois Mangin, PhD4,5; Marie-Odile Habert,
MD PhD4,6,7; Stephanie Bombois, MD PhD8; Pierre-Jean Ousset, MD PhD9; Florence Pasquier, MD
PhD10; Olivier Hanon, MD PhD11; Claire PAQUET, MD PhD12; Audrey GABELLE, MD PhD13;
Mathieu Ceccaldi, MD PhD14; Cédric Annweiler, MD PhD15,16; Pierre Krolak-Salmon, MD PhD17;
Yannick Béjot, MD PhD18; Catherine Belin, MD PhD19; David Wallon, MD PhD20; Mathilde Sauvee,
MD PhD21; Emilie Beaufils, MD PhD22; Isabelle Bourdel-Marchasson, MD PhD23, 24; Isabelle
Jalenques, MD PhD25; Marie Chupin, PhD4,26; Geneviève Chêne, MD PhD1,2; Carole Dufouil, PhD1,2 on behalf of the MEMENTO cohort Study Group
This is an open access article distributed under the terms of the Creative Commons
Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which
permits downloading and sharing the work provided it is properly cited. The work
cannot be changed in any way or used commercially without permission from the
journal.
Neurology® Published Ahead of Print articles have been peer reviewed and accepted for
publication. This manuscript will be published in its final form after copyediting, page
composition, and review of proofs. Errors that could affect the content may be corrected
during these processes.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
Published Ahead of Print on July 1, 2021 as 10.1212/WNL.0000000000012440
Equal Author Contributions: Geneviève Chêne and Carole Dufouil cntributed equally to this work as senior co-authors
Corresponding Author: Carole Dufouil [email protected]
Affiliation Information for All Authors: 1. Univ. Bordeaux, Inserm, UMR 1219, Inserm, CIC1401-EC, F-33000 Bordeaux, France; 2. Pole de sante publique Centre Hospitalier Universitaire (CHU) de Bordeaux, F-33000 Bordeaux, France;3. Univ. Bordeaux, Inserm, UMR 1219, F-33000 Bordeaux, France; 4. CATI Multicenter Neuroimaging Platform, F-75000 Paris, France;5. Neurospin CEA Paris Saclay University, F-91190 Gif-sur-Yvette, France; 6. Sorbonne Université, CNRS, INSERM, Laboratoire d’Imagerie Biomédicale, LIB, F-75006, Paris, France;7. AP-HP, Hôpital Pitié-Salpêtrière, Médecine Nucléaire, F-75013, Paris, France; 8. IM2A AP-HP INSERM UMR-S975 Groupe Hospitalier Pitié-Salpêtrière Institut de la Mémoire et de la Maladie d'Alzheimer Institut du Cerveau et de la Moelle épinière Sorbonne Université Paris, France;9. Inserm UMR1027, Université de Toulouse III Paul Sabatier, F-31000 Toulouse, France; 10. Univ Lille, Inserm 1171, CHU, Centre Mémoire (CMRR) Distalz, F-59000 Lille, France;11. Service de Gériatrie, Université Paris Descartes, Hôpital Broca, F-75013 Paris, France; 12. Université de Paris, Centre de Neurologie Cognitive Hôpital Lariboisière, INSERMU1144, F-75010, Paris, France; 13. Clinical and Research Memory center of Montpellier, Department of Neurology, Gui de Chauliac Hospital, University of Montpellier, Inserm U1061, F-34000 Montpellier, France; 14. CMMR PACA Ouest CHU Timone APHM & Aix Marseille Univ INSERM INS Inst Neurosci Syst, F-13000, Marseille, France; 15. Department of Geriatric Medicine, Angers University Hospital, Angers University Memory Clinic, Research Center on Autonomy and Longevity, UPRES EA 4638, University of Angers, F-49000 Angers, France; 16. Robarts Research Institute, Department of Medical Biophysics, Schulich School of Medicine and Dentistry, the University of Western Ontario, London, ON, Canada. 17. Univ. Lyon, Inserm U1028, CNRS UMR5292, Centre de Recherche en Neurosciences de Lyon, Centre Mémoire Ressource et Recherche de Lyon (CMRR), Hôpital des Charpennes, Hospices Civils de Lyon, F-69000 Lyon, France;18. Univ. Bourgogne, EA7460, Centre Mémoire de Ressources et de Recherches, CHU Dijon Bourgogne, F-21000 Dijon, France; 19. Service de Neurologie Hôpital Saint-Louis AP-HP, F-75010 Paris, France;20. Univ. Normandie, UNIROUEN, Inserm U1245, Departement de Neurologie, CNR-MAJ, CHU de Rouen, F-76000 Rouen, France; 21. CMRR Grenoble Arc Alpin, CHU Grenoble, F-38000 Grenoble, France;22. CMRR, University Hospital Tours, F-37000 Tours, France; 23. Centre de Résonance Magnétique des Systèmes Biologiques, UMR 5536 Université de Bordeaux/CNRS, F-33000, Bordeaux, France;24. Pole de gérontologie clinique CHU de Bordeaux, F-33000 Bordeaux, France; 25. Memory Resource and Research Centre of Clermont-Ferrand, CHU de Clermont-Ferrand, and Clermont Auvergne University, F-63000 Clermont-Ferrand, France;26. Institut du Cerveau et de la Moelle épinière, Inserm, U 1127,3 CNRS, UMR 7225, Sorbonne Université, CATI, F-75013, Paris, France;
Contributions: Eric Frison: Drafting/revision of the manuscript for content, including medical writing for content; Study concept or design; Analysis or interpretation of data
Cecile Proust-Lima: Drafting/revision of the manuscript for content, including medical writing for content; Study concept or design; Analysis or interpretation of data
Jean-Francois Mangin: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data; Study concept or design
Marie-Odile Habert: Drafting/revision of the manuscript for content, including medical writing for
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
content; Major role in the acquisition of data; Study concept or design
Stephanie Bombois: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Pierre-Jean Ousset: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Florence Pasquier: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Olivier Hanon: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Claire PAQUET: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Audrey GABELLE: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Mathieu Ceccaldi: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Cédric Annweiler: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Pierre Krolak-Salmon: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Yannick Béjot: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Catherine Belin: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
David Wallon:Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Mathilde Sauvee: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Emilie Beaufils: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Isabelle Bourdel-Marchasson: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Isabelle Jalenques: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data
Marie Chupin: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data; Study concept or design
Geneviève Chêne: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data; Study concept or design; Analysis or interpretation of data
Carole Dufouil: Drafting/revision of the manuscript for content, including medical writing for content; Major role in the acquisition of data; Study concept or design; Analysis or interpretation of data
Number of characters in title: 73
Abstract Word count: 215
Word count of main text: 3394
References: 46
Figures: 1
Tables: 5
Supplemental: STROKE
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
Statistical Analysis performed by: Eric Frison, MD PhD, Bordeaux University Hospital
Search Terms: [ 26 ] Alzheimer's disease, [ 36 ] Cognitive aging, [ 54 ] Cohort studies, [ 120 ] MRI, [ 122 ] PET
Acknowledgements: The MEMENTO cohort is sponsored by Bordeaux University Hospital (coordination: CIC1401-EC, Bordeaux) and was funded through research grants from the Fondation Plan Alzheimer (Alzheimer Plan 2008–2012), the French ministry of research and higher education (Plan Malandies Neurodégénératives (2016-2020)). The MEMENTO cohort has received funding support from AVID, GE Healthcare, and FUJIREBIO through private-public partnerships. The Insight-PreAD sub-study was promoted by INSERM in collaboration with the Institut du Cerveau et de la Moelle épinière, Institut Hospitalo-Universitaire, and Pfizer and has received support within the “Investissement d'Avenir” (ANR-10-AIHU-06) program. Sponsor and funders were not involved in the study conduct, analysis and interpretation of data.
Study Funding: The authors report no targeted funding
Disclosures: E Frison, C. Proust-Lima, JF Mangin, MO Habert, S Bombois, PJ Ousset, Florence Pasquier, Olivier Hanon, Claire Paquet, A Gabelle, M Ceccaldi, C Annweiler P Krolak-Salmon, Y Béjot, C Belin, D Wallon, M Sauvée, E Beaufils, I Bourdel-Marchasson, I Jalenques, M Chupin, G Chêne, C Dufouil report no disclosures relevant to the manuscript.
Appendix 2-http://links.lww.com/WNL/B459
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
ABSTRACT
OBJECTIVE: To assess the role of biomarkers of Alzheimer’s Disease (AD),
neurodegeneration and small vessel disease (SVD) as mediators in the association
between diabetes mellitus and cognition.
METHODS: The study sample was derived from MEMENTO, a cohort of French
adults recruited in memory clinics and screened for either isolated subjective
cognitive complaints or mild cognitive impairment. Diabetes was defined based on
blood glucose assessment, use of antidiabetic agent or self-report. We used
structural equation modelling to assess whether latent variables of AD pathology
(PET mean amyloid uptake, Aβ42/Aβ40 ratio and CSF phosphorylated tau), SVD
(white matter hyperintensities volume and visual grading), and neurodegeneration
(mean cortical thickness, brain parenchymal fraction, hippocampal volume, and
mean fluorodeoxyglucose uptake) mediate the association between diabetes and a
latent variable of cognition (five neuropsychological tests), adjusting for potential
confounders.
RESULTS: There were 254 (11.1%) participants with diabetes among 2,288
participants (median age 71.6 years; 61.8% women). The association between
diabetes and lower cognition was significantly mediated by higher
neurodegeneration (standardized indirect effect: -0.061, 95% confidence interval: -
0.089; -0.032), but not mediated by SVD and AD markers. Results were similar when
considering latent variables of memory or executive functioning.
CONCLUSION: In a large clinical cohort in the elderly, diabetes is associated with
lower cognition through neurodegeneration, independently of SVD and AD
biomarkers.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
INTRODUCTION
Type 2 diabetes (diabetes) is a risk factor for cognitive decline and dementia (1,2).
Several underlying mechanisms could be involved, such as chronic hyperglycemia
leading to advanced glycation end-products, atherosclerosis, and subsequent
cerebrovascular lesions (3–5). Insulin dysregulation, including insulin resistance and
insulin deficiency, may promote cerebral hypometabolism (6) and amyloid and tau
pathologies, hallmarks of Alzheimer’s disease (AD) (7). Diabetes has also been
associated with brain structural modifications such as cerebral atrophy and
cerebrovascular lesions (8–10). Moreover, while diabetes is associated with cerebral
hypometabolism (11,12), results are conflicting regarding its association with amyloid
and tau pathology, whether measured in the brain (PET) or in CSF (11,13,14).
Previous studies have suggested a mediating role of neurodegeneration and small
vessel disease biomarkers on the association between diabetes and cognition (15–
17). However, the mediating role of AD-specific lesions (amyloid plaques and
neurofibrillary tangles), and the correlation between those different brain features
have not been considered so far.
We thus estimated the mediating effect of biomarkers of AD, neurodegeneration and
small vessel disease in the association between diabetes and cognition, in non-
demented older adults recruited from French memory clinics.
METHODS
The MEMENTO Cohort
The MEMENTO cohort is a clinic-based study of patients presenting with a large
variety of cognitive symptoms or subjective cognitive complaints, who were enrolled
between April 2011 and June 2014, within the French national network of university
hospital-based memory clinics (18). At inclusion, participants presented either 1) with
mild cognitive impairment, when performing one standard deviation worse than the
mean of the subject’s own age, sex, and education-level group, in one or more
cognitive domains, this deviation being identified for the first time through cognitive
tests performed recently (less than 6 months preceding screening phase), or 2) with
isolated cognitive complaints, if participants had subjective cognitive complaint
(assessed through visual analogic scale), without any objective cognitive deficit as
defined previously, while being 60 years and older. All participants had a Clinical
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
Dementia Rating scale (19) score ≤0.5. Main exclusion criteria have been described
elsewhere (22). All examinations (including neuropsychological battery
administration, clinical examinations, brain MRI, CSF samples and
fluorodeoxyglucose [FDG] and amyloid PET) followed standardized procedures (18).
Among the 2,323 participants included in the MEMENTO cohort, 2,288 participants
from 26 study centers were included in this analysis after exclusion of participants
with missing data on diabetes status (N = 35).
Standard protocol approvals, registrations, and patient consents
This study was performed in accordance with the Declaration of Helsinki. All
participants provided written informed consent. The MEMENTO cohort protocol has
been approved by the local ethics committee (“Comité de Protection des Personnes
Sud-Ouest et Outre Mer III”; approval number 2010-A01394-35) and was registered
in ClinicalTrials.gov (Identifier: NCT01926249).
Diabetes definition
Participants were classified as having diabetes at baseline visit either in presence of
fasting blood glucose ≥ 7 mmol/L (≥126 mg/dL) or non-fasting blood glucose ≥ 11.1
mmol/L (≥200 mg/dL) or antidiabetic drug intake (Anatomical Therapeutic Chemical
classification system: code A10A “insulins and analogues”, and code A10B “blood
glucose lowering drugs, excl. insulins”) or self-reported history of diabetes.
Neuropsychological evaluation
A full neuropsychological test battery was administered to participants (18). Global
cognition was assessed by Mini-Mental State Examination (MMSE) (20), long-term
memory was assessed by Free and Cued Selective Reminding Test (FCSRT) (21),
semantic verbal fluency via ‘animal’ words (22), visuo-spatial abilities by Rey-
Osterrieth Complex Figure Test (23), and attention and executive functions by Trail
Making Test (TMT) A and B (24).
Biomarkers assessment
MRI
As part of the inclusion criteria, participants had to agree to undergo brain MRI. Brain
magnetic resonance images were acquired after a standardization of the imaging
processes and coordinated by the CATI (http://cati-neuroimaging.com), a
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
neuroimaging platform dedicated to multicentre studies (25). Full details are
described elsewhere (18). Briefly, MRI machines of 1.5 and 3 Tesla were used
across centers using harmonized protocols. All MRI scans acquired were then
centralized, quality checked, and postprocessed to obtain standardized
measurements for each participant. Whole-brain, gray matter, and white matter
volumes were assessed with Statistical Parametric Mapping 8 (26), hippocampal
volumes with the SACHA software (27), and mean cortical thickness of each
hemisphere with FreeSurfer 5.3 averaged in the ROI of the Desikan-Killiany atlas
(28). White matter lesions volumetry was performed using WHASA software (29)
complemented by a centralized visual assessment by a trained rater using the
Fazekas and Schmidt scale (30).
FDG-PET
18F-FDG-PET was offered to all participants but was not mandatory. PET images
were acquired after a standardization of the acquisition and reconstruction imaging
parameters, coordinated by the CATI (31). After a centralized quality check and
postprocessing performed by the CATI, the following measures were obtained: mean
FDG-PET uptake for the regions of interest (ROIs) of the Automated Anatomical
Labeling atlas relative to the pons reference region (32), including partial volume
correction, and mean FDG-PET uptake for a set of AD-specific ROIs inferred from
the Alzheimer's Disease Neuroimaging Initiative database (33), expressed as
standard uptake value ratios (SUVRs).
PET amyloid imaging
PET amyloid imaging was available for 643 participants of the analytical sample,
using either 18F-florbetapir (Amyvid®, Eli Lilly) (N=437) or 18F-flutemetamol
(Vizamyl®, GE Healthcare) (N=206) radioligands. Mean brain amyloid SUVR was
computed, harmonized across the radioligands (34), and used for the current study.
CSF sampling
Lumbar puncture was offered to all participants but was not mandatory , and CSF
centralized measurements of amyloid-β 42 peptide (Aβ42), Aβ40, total tau, and
phosphorylated tau levels were performed using the standardised INNOTEST
sandwich ELISA (Fujirebio, Ghent, Belgium).
Potential confounding factors
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
Sociodemographic information recorded at baseline included age, sex and education
(low education defined as no or primary school, intermediate education defined as
secondary school or high school, and high education defined as university). Lifestyle
factors included smoking status (never, former and current smoker) and current
alcohol consumption (no, ≤ 1 drink/day, and >1 drink/day). Hypertension was defined
as antihypertensive drug intake or mean of three blood pressure measurements
either ≥ 140 mmHg for systolic blood pressure or ≥ 90 mmHg for diastolic blood
pressure. Dyslipidemia was defined by plasma cholesterol > 6.24 mmol/L or use of
any lipid-lowering drugs. Body mass index (BMI) was categorized as <20 kg/m², 20
to 25 kg/m², 25.1-29.9kg/m² and ≥30kg/m². History of cardiovascular disease was
defined as a self-reported history of myocardial infarction, angina pectoris, coronary
artery, or peripheral artery disease. History of stroke was self-reported. Depression
was assessed with the Neuropsychiatric Inventory–Clinician (NPI-C) (35). APOE ε2,
ε3, or ε4 alleles were determined for all participants by KBiosciences (Hoddesdon,
UK; www.kbioscience.co.uk) as described elsewhere (18). APOE ε4 status was
defined as presence of at least one ε4 allele versus absence.
Statistical analyses
Baseline characteristics were compared according to baseline diabetic status for the
analytical sample. We used chi-square test (or Fisher exact test when appropriate)
and Student t test (or non-parametric Mann-Whitney-Wilcoxon test when
appropriate) for categorical and continuous variables comparisons, respectively.
Brain parenchymal fraction was computed as the sum of grey matter and white
matter volumes divided by total intracranial volume. Total hippocampal volume was
computed as the sum of left and right hippocampal volumes. WMH volume and
hippocampal volume were adjusted for total intracranial volume using the residual
approach (36). Mean FDG uptake across the brain was used.
Structural equation modeling (SEM) (37) was used to examine a potential mediating
role of biomarkers respectively of AD, small vessel disease (SVD) and
neurodegeneration in the association between diabetes and cognition. SEM was
preferred over standard regression modeling for its ability to directly focus the
mediation analysis on the dimensions of interest (here cognition, SVD, AD and
neurodegeneration), and to define each dimension from several noisy observed
indicators. The observed indicators of the four latent variables of interest, namely AD
pathology, small vessel disease, neurodegeneration and cognition, are listed in
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
Table 1 . They were determined from the literature and validated in preliminary
separated SEM analyses. Correlated residuals were assumed between left and right
cortical thicknesses and between TMT A and TMT B scores to account for a
potential common source of measurement error. Mean brain amyloid SUVR was
normalized using a logarithmic transformation and then standardized (z-score) by
radioligand. The relationships between diabetes, potential confounders, and latent
variables of AD pathology, neurodegeneration, small vessel disease, and cognition
were modelled in the structural linear regressions. For ease of interpretation, the four
latent variables were standardized (mean 0, variance 1) so that one unit corresponds
to the standard deviation of a given dimension. The indirect effects of diabetes on
cognition through the latent dimensions were estimated with their 95% CI, using path
analysis technique (37). All linear regressions of mediators and cognition were
adjusted for the following potential confounding factors: age, sex, education (high
education versus low and intermediate), smoking status (current smoker versus
never or former smoker), alcohol consumption (>1 drink/day versus ≤1 drink/day),
hypertension, dyslipidemia, obesity (≥30kg/m²) and APOE genotype (ε4 carrier
versus ε4 non-carrier). Missing values for observed indicators of latent variables and
for confounding factors were handled using a full information maximum likelihood
approach, assuming missingness at random. The multicentric nature of the data was
accounted for and Huber-White robust standard errors were reported to correct for
the potential intra-center correlation (38). The general goodness of fit was evaluated
using robust Tucker-Lewis Index (TLI), robust Comparative Fit Index (CFI), robust
Root Mean Square Error of Approximation (RMSEA) and its 90% confidence interval,
p-value for test of close fit (null hypothesis RMSEA <0.05), and Standardized Root
Mean Square Residual (SRMR) with cut-offs recommended in the literature (39).
Several sensitivity analyses were performed. First, we used a different definition of
“diabetes” by excluding a self-reported history of diabetes. Second, additional
baseline characteristics associated with availability of MRI, FDG-TEP, amyloid-PET
and CSF data (living alone, Clinical Dementia Rating scale score, prevalent
dementia, depression, stroke history, cardiovascular history, and physical activity
expressed as metabolic equivalent of task minutes per week, Table 2 ) were used as
auxiliary variables in the estimation process under FIML to strengthen the missing at
random assumption. Third, as the mediation analysis framework makes the implicit
assumption that mediators (i.e., AD pathology, small vessel disease and
neurodegeneration) are anterior to the outcome (i.e., cognition), we tried to preserve
this assumption by excluding biomarkers measurements performed more than 6
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
months after cognitive assessments. Fourth, as CSF biomarkers are prone to
variability whereas brain biomarkers are indicators of accumulated burden of lesions
(40), we performed a sensitivity analysis using only brain amyloid load as indicator of
the latent variable for AD pathology. Finally, we also compared the results with those
obtained when considering interactions between diabetes and each mediator in the
main adjusted model, as recommended for mediation analysis (41).
We also explored the mediating pathways in the association of diabetes with specific
cognitive domains in separate models: a latent variable for memory (indicators: total
free recall score and verbal fluency) and a latent variable for executive functioning
(indicators: TMT A and TMT B scores).
Analyses were conducted using SAS v9.3 (SAS Institute Inc, Cary, NC, USA), and R
version 3.5.1 (42) with the lavaan package for SEM analysis (38).
Data Availability
Anonymized data will be shared by request from any qualified investigator for the
sole purpose of replicating procedures and results presented in the article and as
long as data transfer is in agreement with EU legislation on the general data
protection regulation.
RESULTS
Baseline description
Compared to participants without diabetes at baseline, participants with diabetes
(254, 11.1%) were more likely to be men, and to have lower education level. They
were also more likely to have hypertension, dyslipidemia, obesity, and history of
cardiovascular disease or stroke. Participants with diabetes had on average lower
performances on executive functions and attention, memory and semantic verbal
fluency (Table 3 ).
At baseline, 65.3% of participants with diabetes were taking antidiabetic medications
(oral antidiabetic agents, 57.5%; insulin, 13.8%). Diabetes status was solely based
on self-report in 60 (23.6%) of the diabetic participants. The median self-reported
duration of diabetes was 10.0 years (interquartile range, 4.9-19.4 years).
Diabetes, latent biomarkers and latent cognition
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
The model fit was adequate according to the recommended cutoffs: robust CFI =
0.951, robust TLI = 0.926, robust RMSEA = 0.040 (90% CI, 0.037; 0.042), p-value
for test of close fit = 1.00, and SRMR = 0.038. Associations between diabetes, AD
pathology, SVD, neurodegeneration and cognition are presented in Figure 1 .
Presence of diabetes was significantly associated with higher neurodegeneration but
was not significantly associated with AD pathology and SVD. Higher levels of small
vessel disease, neurodegeneration and AD pathology were independently
associated with lower cognition. Once adjusted for neurodegeneration, AD pathology
and SVD, there was no direct effect of diabetes on cognition (standardized β =
0.023, 95% CI: -0.030; 0.076, p = 0.40). Association between diabetes and lower
cognition was mainly mediated by higher neurodegeneration (standardized β = -
0.061, 95% CI: -0.089; -0.032, p < 0.001). The indirect effect of diabetes on cognition
via SVD and AD pathology were non-statistically significant (standardized β = 0.000,
95% CI: -0.004; 0.004, p = 0.98 and standardized β = -0.013, 95% CI: -0.040; 0.015,
p = 0.38, respectively).
In complementary analyses considering specific cognitive functions, associations
between diabetes and lower memory or lower executive functioning were also mainly
mediated by higher neurodegeneration (standardized β = -0.058, 95% CI: -0.088; -
0.029, p<0.001 and standardized β = -0.034, 95% CI: -0.051; -0.016, p<0.001
respectively) (Table 4 ).
Sensitivity analyses
Results were similar when excluding self-reported history from the definition of
diabetes, when adding auxiliary variables to the estimation process or when
excluding delayed measures of biomarkers (Table 5 ). When using only brain amyloid
load as indicator of the latent variable for AD pathology, the indirect pathway linking
diabetes to lower cognition through higher neurodegeneration was of similar
magnitude (standardized β = -0.066, 95% CI: -0.097; -0.034, p<0.001). Diabetes was
significantly associated with higher AD pathology (standardized β = 0.107, 95% CI:
0.021; 0.193, p = 0.01), and higher AD pathology was significantly associated with
lower cognition (standardized β = -0.144, 95% CI: -0.248; -0.039, p = 0.007). The
indirect pathway linking diabetes to lower cognition through AD pathology remained
non-statistically significant (standardized β = -0.015, 95% CI: -0.033; 0.002, p = 0.08)
though. When considering interaction between diabetes and each intermediate latent
variable, the indirect effects of diabetes on cognition via neurodegeneration
(standardized β = -0.059, 95%CI: -0.089; -0.030, p < 0.001), AD pathology
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
(standardized β = -0.011, 95%CI: -0.034; 0.012, p = 0.34) and SVD (standardized β
= -0.001, 95%CI: -0.006; 0.003, p = 0.54) remained virtually the same.
DISCUSSION
In a cross-sectional analysis of a large clinical cohort of participants with either
isolated cognitive complaints or mild cognitive impairment, we report that the
deleterious effect of diabetes on cognitive performances is mainly mediated through
markers of neurodegeneration whereas AD pathology (amyloid, p-Tau) or small
vessel disease pathology do not seem to play a major role.
The association between diabetes and markers of neurodegeneration such as brain
atrophy (8,12,13,43) and brain hypometabolism (11,12) has been consistently
reported in cross-sectional studies. While diabetes is a risk factor for vascular
disease and stroke, its association with subclinical cerebrovascular lesions (silent
brain infarcts, WMH, cerebral microbleeds) is uncertain (44). In the present study,
diabetes was not associated with small vessel disease, even though participants with
diabetes had more frequent self-reported history of stroke.
The mediating role of neurodegeneration and small vessel disease in the association
between diabetes and cognition has already been investigated in several studies. In
a sample of 4,206 older adults of the Age, Gene/Environment Susceptibility–
Reykjavik Study (mean age 76 years, 11% with diabetes), MRI markers of
neurodegeneration (gray matter, normal white matter, and total brain tissue volumes)
and small vessel disease (cortical infarcts, subcortical infarcts, WMLs, and CMBs)
significantly mediated the cross-sectional association of diabetes with lower
processing speed and executive function (15). In a longitudinal analysis on 817
participants from the Alzheimer’s Disease Neuroimaging Initiative cohort (mean age
75 years, 15% with diabetes) the effect of diabetes on cognitive decline up to 60
months (mean follow-up time, 30 months) was significantly mediated by baseline
cortical thickness (17). Similarly, in a sample of 448 older adults of the Swedish
National Study on Aging and Care in Kungsholmen (mean age at baseline, 72
years), a higher cardiovascular burden, including diabetes as a component, was
associated with a faster MMSE decline over 9 years; this effect being largely
mediated by brain MRI markers of atrophy (volumes of total gray matter, ventricles,
and hippocampus) and small vessel disease (volume of WMHs) (16). Nevertheless,
none of those studies accounted for AD biomarkers, unlike the present study.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
Insulin resistance and associated insulin signaling impairment promote Aβ
accumulation and tau phosphorylation (7). However, no association between
diabetes and amyloid and tau biomarkers was reported in previous studies
(11,13,45). In the present study, diabetes was associated with higher brain amyloid
load measured on PET imaging, but diabetes was not associated with the latent
variable of AD pathology, which included CSF biomarkers of amyloid and tau. This
discrepancy between brain and CSF biomarkers can partly be explained by the
variability of CSF biomarkers, whereas brain biomarkers are indicators of
accumulated lesions.
Although it needs to be replicated in longitudinal studies, our finding that
neurodegeneration mediates the association between diabetes and cognitive
performances, independently of biomarkers of AD and small vessel disease supports
the hypothesis of a direct role of diabetes-related insulin resistance in the
development of cognitive impairment in older adults with diabetes. Indeed, insulin
also plays an important role in neuronal synaptic plasticity and facilitates learning
and memory in humans (4) and, therefore, impaired insulin signaling could directly
contribute to neuronal dysfunction and degeneration. As impaired insulin signaling
has also been linked to promotion of amyloid-β accumulation and tau
hyperphosphorylation (7), brain insulin resistance could be a therapeutic target in AD
and related dementias. Several exploratory clinical trials have reported a beneficial
effect on cognition of intranasal insulin for healthy participants, participants with
diabetes, mild cognitive impairment or AD (46), and longer-term trials are currently
ongoing.
The MEMENTO study has several strengths to answer the current objectives. First, a
wide range of biomarkers was acquired in a highly standardized setting on more than
2,000 participants allowing a multi-dimensional assessment of brain ageing and
pathology biomarkers. Indeed, we were able to include simultaneously brain MRI,
brain FDG-PET, amyloid-PET and CSF data in a mediation analysis of the diabetes-
cognition association, offering a unique insight on underlying mechanisms. Second,
we were able to model brain biomarkers as latent variables in a SEM framework,
accounting for measurement error of the indicators, and we were able to estimate
direct and indirect effects of diabetes on several domains of cognition. Third, results
were robust to several sensitivity analyses. There are also some limitations. First, the
temporal relationship between diabetes, biomarkers and cognition is not ensured by
the cross-sectional design, and causality cannot be claimed. Nevertheless, we can
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
hypothesize that diabetes preceded biomarkers measures in most participants with
diabetes (duration was 4.9 years or more in 75% of participants with diabetes). We
also modeled correlations between neurodegeneration, AD pathology and SVD
instead of directed relationships because the causal interpretation of their
interrelations requires longitudinal data. Second, no tau-PET data was available to
assess tau pathology, and we had to use CSF phosphorylated tau as a proxy for
cerebral tau accumulation, assuming a strong correlation between both, as
suggested by existing evidence (40). Third, the analytical strategy relies on the
assumption that data are missing at random. This assumption may be strong for
CSF and PET-amyloid data, for which 70% to 80% of data were missing. However,
we used a broad range of baseline characteristics associated with availability of CSF
and PET-amyloid data as auxiliary variables in the estimation process, thus making
the missing-at-random assumption more plausible. We must also acknowledge the
unavailability of data regarding past and current glucose control that prevented us to
explore whether diabetes control modified the explored relationships. Finally, the
observed findings may not fully translate in the general older population, as
participants in the MEMENTO study are adults with either isolated cognitive
complaints or mild cognitive impairment who were seeking care in memory clinics.
The current results suggest that the detrimental effect of diabetes on cognition is
mediated by neurodegeneration, independently of AD and small vessel disease
pathologies, in a population of older adults at risk for dementia. Longitudinal studies
are now needed to reinforce and confirm these findings.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
TABLES
Table 1. Observed indicators for latent dimensions variables
Latent variables Observed indicators Data available
N (%)
Small vessel disease
White matter hyperintensities volume 1,884 (80.6%)
Fazekas scale scores for paraventricular
white matter hyperintensities 2,145 (93.8%)
Fazekas scale scores for deep white matter
hyperintensities 2,145 (93.8%)
Alzheimer’s disease
pathology
Mean brain amyloid uptake 643 (28.1%)
CSF Aβ42/Aβ40 ratio 400 (17.5%)
CSF Phosphorylated tau 408 (17.8%)
Neurodegeneration
Mean right cortical thickness 2,106 (92.0%)
Mean left cortical thickness 2,106 (92.0%)
Brain parenchymal fraction 2,103 (91.9%)
Hippocampal volume 2,061 (90.1%)
Mean brain FDG uptake 1,308 (57.2%)
Cognition
FCSRT total free recall score 2,269 (99.2%)
TMT A (seconds/correct move) 2,265 (99.0%)
TMT B (seconds/correct move) 2,192 (95.8%)
Rey complex figure test, 3-minute copy score 2,125 (92.9%)
Verbal fluency (number of animals produced) 2,245 (98.1%)
Abbreviations: FCSRT, Free and Cued Selective Reminding Test; TMT, Trail Making Test.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
Table 2. Baseline characteristics associated with the availability of MRI, FDG-PET, amyloid-PET and CSF data – MEMENTO Study, France (n = 2,288).
Available data P a
No Yes
MRI, N 130 2,158
Cardiovascular history 20 (15.4) 185 (8.6) 0.008
MMSE score 27.4 (2.2) 27.9 (1.9) 0.001
FCSRT total free recall score 24.3 (9.2) 26.1 (8.2) 0.01
FDG-PET, N 980 1,308
Female sex 648 (66.1) 765 (58.5) <0.001
Current alcohol consumption 0.006
No 352 (37.1) 399 (30.8)
≤1d/day 412 (43.5) 604 (46.7)
>1d/day 184 (19.4) 291 (22.5)
Dyslipidemia 402 (55.0) 480 (46.3) <0.001
MMSE score 27.8 (2.0) 28.0 (1.9) 0.009
TMT A (seconds/correct move) 2.1 (1.0) 2.0 (0.9) 0.005
TMT B (seconds/correct move) 5.2 (3.6) 4.9 (3.2) 0.02
Rey complex figure test, 3-minute
copy score 14.5 (7.1) 15.6 (6.9) <0.001
Verbal fluency, animals (number of
words produced) 27.7 (8.7) 28.8 (8.7) 0.006
Amyloid -PET, N 1,645 643
Current alcohol consumption <0.001
No 584 (36.4) 167 (26.2)
≤1d/day 713 (44.5) 303 (47.5)
>1d/day 307 (19.1) 168 (26.3)
Diabetes 201 (12.2) 53 (8.2) 0.007
Dyslipidemia 642 (52.8) 240 (43.6) <0.001
Depression 677 (41.2) 212 (33.0) <0.001
Clinical Dementia Rating scale <0.001
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
0 540 (33.0) 383 (59.8)
0.5 1,096 (67.0) 258 (40.2)
MMSE score 27.7 (2.1) 28.3 (1.5) <0.001
FCSRT total free recall score 25.0 (8.6) 28.4 (6.9) <0.001
TMT A (seconds/correct move) 2.1 (1.0) 1.9 (0.7) <0.001
TMT B (seconds/correct move) 5.3 (3.6) 4.5 (2.7) <0.001
Rey complex fig ure test, 3 -minute
copy score 14.7 (7.1) 16.4 (6.6) <0.001
Verbal fluency, animals (number of
words produced) 27.5 (8.8) 30.3 (8.2) <0.001
CSF, N 1,877 411
Age (years) 71.3 (8.6) 68.8 (8.8) <0.001
Female sex 1197 (63.8) 216 (52.6) <0.001
Living alone 602 (32.4) 101 (24.6) 0.002
Physical activity, MET-hour/week 52.2 (47.2) 59.7 (52.9) 0.01
Clinical Dementia Rating scale 0.02
0 777 (41.6) 146 (35.5)
0.5 1089 (58.4) 265 (64.5)
APOE ε4 carrier 501 (28.0) 155 (38.9) <0.001
MMSE 27.9 (1.9) 27.7 (2.0) 0.001
FCSRT total free recall score 26.3 (8.2) 24.6 (8.8) <0.001
Verbal fluency, animals (number of
words produced) 28.4 (8.7) 27.9 (8.9) 0.04
Abbreviations: FCSRT, Free and Cued Selective Reminding Test; MET, metabolic equivalent of task; MMSE, Mini-Mental State Examination; TMT, Trail Making Test. a P-values for comparison using t-tests for quantitative variables and chi-square test or Fisher test for qualitative variables. Comparisons for cognitive tests were adjusted for age, sex and education.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
Table 3. Baseline characteristics according to diabetes – MEMENTO Cohort, France (n = 2,288)
Diabetes
No
(n = 2,034)
Yes
(n = 254) P a
Age (years) 70.9 (8.8) 70.8 (7.9) 0.80
Female sex 1,302 (64.0) 111 (43.7) <0.001
Education 0.02
Low 487 (23.9) 71 (28.0)
Intermediate 722 (35.5) 103 (40.6)
High 823 (40.5) 80 (31.5)
Smoking status 0.05
Never 1,191 (59.0) 137 (54.8)
Former 676 (33.5) 101 (40.4)
Current 151 (7.5) 12 (4.8)
Current alcohol consumption 0.17
No 658 (33.0) 93 (37.8)
Up to 1 drink/day 918 (46.0) 98 (39.8)
>1 drink/day 420 (21.0) 55 (22.4)
Body mass index (kg/m²) <0.001
<20 145 (7.3) 6 (2.4)
20-25 910 (45.7) 68 (27.6)
25.1-29.9 712 (35.8) 92 (37.4)
≥30 223 (11.2) 80 (32.5)
Hypertension 1,135 (59.8) 188 (77.4) <0.001
Dyslipidemia 761 (48.9) 127 (60.5) 0.002
Self-reported cardiovascular history 156 (7.7) 49 (19.3) <0.001
Self-reported stroke history 76 (3.7) 16 (6.3) 0.05
Depression 791 (38.9) 98 (38.6) 0.92
APOE ε4 carrier 596 (30.6) 60 (24.6) 0.05
Cognitive tests
MMSE score 28.0 (1.9) 27.6 (2) 0.03 b
FCSRT total free recall score 26.2 (8.4) 24.2 (7.4) 0.03 b
TMT A (seconds/correct move) 2.05 (0.94) 2.16 (0.88) 0.02 c
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
TMT B (seconds/correct move) 4.97 (3.39) 5.57 (3.41) <0.001 d
Rey complex figure test, 3-minute
copy score 15.1 (7.0) 15.5 (7.0) 0.89 b
Verbal fluency, (number of animals
produced) 28.5 (8.7) 26.9 (8.7) 0.04 b
Missing data: education, 2; smoking status, 20; alcohol consumption, 46; body mass index, 52; hypertension, 148; dyslipidemia, 521; APOE genotype, 98; MMSE, 6; FCSRT, 19; TMT A, 23; TMT B, 96; Rey complex figure, 163; verbal fluency, 43. Abbreviations: FCSRT, Free and Cued Selective Reminding Test; MMSE, Mini-Mental State Examination; TMT, Trail Making Test. a P-values for comparison using t-tests for quantitative variables and chi-square test or Fisher test for qualitative variables, except when stated otherwise b P-values for comparison using linear regression modeling adjusted on age, sex and education. c P-value for comparison of log-transformed values of TMT A, adjusted on age, sex and education. d P-value for comparison of log-transformed values of TMT B, adjusted on age, sex and education.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
Table 4. Association between diabetes, biomarkers of small vessel disease, neurodegeneration and Alzheimer’s disease, and specific cognitive domains – Structural equation model
Latent variable of memory
Latent variable of executive functioning
Standardized
estimate (95% CI)
P Standardized
estimate (95% CI)
P
Direct effect of diabetes on SVD 0.001 (-0.035; 0.037) 0.95 0.001 (-0.034; 0.037) 0.94 AD pathology 0.047 (-0.059; 0.153) 0.38 0.053 (-0.049; 0.155) 0.31 Neurodegeneration 0.108 (0.071; 0.145) <0.001 0.110 (0.074; 0.146) <0.001
Direct effect of
Diabetes on cognition 0.016 (-0.037; 0.069) 0.55 -0.017 (-0.070; 0.036) 0.53
SVD on cognition -0.104 (-0.169; -0.040) 0.001 -0.094 (-0.163; -0.024) 0.008
Neurodegeneration on cognition -0.542 (-0.737; -0.346) <0.001 -0.306 (-0.441; -
0.171) <0.001
AD pathology on cognition -0.282 (-0.421; -0.144) <0.001
-0.169 (-0.269; -0.068) 0.001
Correlation between
SVD and AD pathology 0.159 (0.064; 0.253) <0.001 0.151 (0.057; 0.245) 0.001 SVD and neurodegeneration 0.038 (-0.056; 0.133) 0.42 0.023 (-0.077; 0.123) 0.65 AD and neurodegeneration 0.257 (0.116; 0.398) <0.001 0.256 (0.128; 0.384) <0.001
Indirect effect of diabetes on cognition
Through SVD 0.000 (-0.004; 0.004) 0.95 0.000 (-0.003; 0.003) 0.94 Through AD pathology -0.013 (-0.042; 0.015) 0.36 -0.009 (-0.027; 0.010) 0.34
Through neurodegeneration -0.058 (-0.088; -0.029) <0.001 -0.034 (-0.051; -0.016) <0.001
Model fit indices Robust CFI 0.963 0.974 Robust TLI 0.937 0.956 Robust RSMEA (90% CI) 0.038 (0.035; 0.041) 0.032 (0.029; 0.035) p-value for test of close fit 1.00 1.00 SRMR 0.035 0.035
Abbreviations: AD, Alzheimer’s disease; CFI, comparative fit index; RSMEA, root mean square error of
approximation; SRMR, Standardized Root Mean Square Residual; SVD, small vessel disease; TLI, Tucker-Lewis
Index.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
Table 5. Association between diabetes, biomarkers and global cognition - Sensitivity analyses
Excluding self-reported
history of
diabetes
Adding
auxiliary
variables
Excluding delayed
biomarker
measurements
(>6months)
Using only
brain
biomarkers
as indicators
Standardized
estimate
(95% CI)
P
Standardized
estimate
(95% CI)
P
Standardized
estimate
(95% CI)
P
Standardized
estimate
(95% CI)
P
Direct effect of
diabetes on
SVD -0.006
(-0.045; 0.033) 0.77
0.002
(-0.035; 0.038) 0.92
0.006
(-0.031; 0.043) 0.75
0.001
(-0.035; 0.036) 0.97
AD pathology 0.049
(-0.046; 0.143) 0.31
0.044
(-0.067; 0.155) 0.44
-0.007
(-0.172; 0.159) 0.94
0.107
(0.021; 0.193) 0.01
Neurodegeneration 0.084
(0.049; 0.121) <0.001
0.109
(0.072; 0.146) <0.001
0.106
(0.068; 0.144) <0.001
0.108
(0.071; 0.144) <0.001
Direct effect on
cognition of
Diabetes 0.030
(-0.017; 0.076) 0.21
0.023
(-0.030; 0.077) 0.39
0.012
(-0.047; 0.072) 0.69
0.030
(-0.020; 0.080) 0.23
SVD -0.114
(-0.185; -0.044) <0.001
-0.113
(-0.183; -0.043) 0.001
-0.108
(-0.187; -0.029) 0.007
-0.131
(-0.201; -0.061) <0.001
Neurodegeneration -0.576
(-0.743; -0.408) 0.001
-0.565
(-0.731; -0.399) <0.001
-0.601
(-0.765; -0.436) <0.001
-0.609
(-0.777; -0.442) <0.001
AD pathology -0.273
(-0.391; -0.154) <0.001
-0.275
(-0.403; -0.147) <0.001
-0.285
(-0.459; -0.111) 0.001
-0.144
(-0.248; -0.039) 0.007
Correlation
between
SVD and
AD pathology
0.157
(0.060; 0.254) 0.002
0.163
(0.066; 0.260) <0.001
0.161
(0.025; 0.298) 0.02
0.156
(0.054; 0.257) 0.003
SVD and
neurodegeneration
0.040
(-0.053; 0.134) 0.39
0.038
(-0.133; 0.057) 0.43
0.039
(-0.056; 0.134) 0.42
0.039
(-0.056; 0.133) 0.42
AD and
neurodegeneration
0.269
(0.130; 0.409) <0.001
0.259
(0.127; 0.390) <0.001
0.160
(0.004; 0.316) 0.04
0.236
(0.096; 0.376) 0.001
Indirect effect of
diabetes on
cognition
Through SVD 0.001
(-0.004; 0.005) 0.77
0.000
(-0.004; 0.004) 0.92
-0.001
(-0.005; 0.003) 0.75
0.000
(-0.005; 0.005) 0.97
Through AD
pathology
-0.013
(-0.037; 0.011) 0.28
-0.012
(-0.042; 0.018) 0.42
0.002
(-0.046; 0.049) 0.93
-0.015
(-0.033; 0.002) 0.08
Through
neurodegeneration
-0.048
(-0.075; -0.021) <0.001
-0.061
(-0.091; -0.032) <0.001
-0.064
(-0.093; -0.035) <0.001
-0.066
(-0.097; -0.034) <0.001
Model fit indices
Robust CFI 0.951 0.951 0.948 0.953
Robust TLI 0.926 0.926 0.921 0.924
Robust RSMEA
(90% CI)
0.040
(0.037; 0.042)
0.040
(0.037 ; 0.042)
0.040
(0.038 ; 0.043)
0.043
(0.040 ; 0.046)
p-value for test
of close fit 1.00 1.00 1.00 1.00
SRMR 0.038 0.032 0.042 0.033
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
Abbreviations: AD, Alzheimer’s disease; CFI, comparative fit index; RSMEA, root mean square error of
approximation; SRMR, Standardized Root Mean Square Residual; SVD, small vessel disease; TLI, Tucker-Lewis
Index.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
FIGURE LEGEND
Figure 1. Structural equation model for the association between diabetes, small vessel disease, neurodegeneration, Alzheimer’s disease biomarkers and cognition
Latent variables of interest are indicated in ovals and observed variables in rectangles. Directed arrows represent linear regressions. Bidirectional arrows represent correlations. Standardized regression coefficients estimates are presented with their 95% confidence interval. Solid lines indicate statistically significant associations and correlations at the 5% level. Dotted lines indicate non-statistically significant associations and correlations at the 5% level. Adjustment covariates and their directed arrows to small vessel disease, neurodegeneration, Alzheimer’s disease biomarkers and cognition are represented in grey. For readiness, the observed indicators defining each latent variable (listed in Table 1) and residual variances for all variables were omitted. AD, Alzheimer’s disease. * p<0.001
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
References
1. Chatterjee S, Peters SAE, Woodward M, Arango SM, Batty GD, Beckett N, et al. Type 2 Diabetes as a Risk Factor for Dementia in Women Compared With Men: A Pooled Analysis of 2.3 Million People Comprising More Than 100,000 Cases of Dementia. Diabetes Care. 2016;39(2):300–7.
2. Rawlings AM, Sharrett AR, Albert MS, Coresh J, Windham BG, Power MC, et al. The Association of Late-Life Diabetes Status and Hyperglycemia With Incident Mild Cognitive Impairment and Dementia: The ARIC Study. Diabetes Care. 2019;42(7):1248–54.
3. Mayeda ER, Whitmer RA, Yaffe K. Diabetes and Cognition. Clin Geriatr Med. 2015;31(1):101–ix.
4. Verdile G, Fuller SJ, Martins RN. The role of type 2 diabetes in neurodegeneration. Neurobiol Dis. 2015;84:22–38.
5. Biessels GJ, Despa F. Cognitive decline and dementia in diabetes mellitus: mechanisms and clinical implications. Nat Rev Endocrinol. 2018;14(10):591–604.
6. Willette AA, Bendlin BB, Starks EJ, et al. Association of insulin resistance with cerebral glucose uptake in late middle–aged adults at risk for Alzheimer Disease. JAMA Neurol. 2015;72(9):1013–20.
7. Bharadwaj P, Wijesekara N, Liyanapathirana M, Newsholme P, Ittner L, Fraser P, et al. The Link between Type 2 Diabetes and Neurodegeneration: Roles for Amyloid-β, Amylin, and Tau Proteins. J Alzheimer’s Dis. 2017;59(2):421–32.
8. Moran C, Phan TG, Chen J, Blizzard L, Beare R, Venn A, et al. Brain atrophy in type 2 diabetes: regional distribution and influence on cognition. Diabetes Care. 2013;36(12):4036–42.
9. Schneider ALC, Selvin E, Sharrett AR, Griswold M, Coresh J, Jack CR, et al. Diabetes, Prediabetes, and Brain Volumes and Subclinical Cerebrovascular Disease on MRI: The Atherosclerosis Risk in Communities Neurocognitive Study (ARIC-NCS). Diabetes Care. 2017;40(11):1514–21.
10. Marseglia A, Fratiglioni L, Kalpouzos G, Wang R, Bäckman L, Xu W. Prediabetes and diabetes accelerate cognitive decline and predict microvascular lesions: A population-based cohort study. Alzheimers Dement. 2019;15(1):25–33.
11. Roberts RO, Knopman DS, Cha RH, Mielke MM, Pankratz VS, Boeve BF, et al. Diabetes and elevated hemoglobin A1c levels are associated with brain hypometabolism but not amyloid accumulation. J Nucl Med. 2014;55(5):759–64.
12. Li W, Risacher SL, Huang E, Saykin AJ, Alzheimer’s Disease Neuroimaging Initiative. Type 2 diabetes mellitus is associated with brain atrophy and hypometabolism in the ADNI cohort. Neurology. 2016;87(6):595-600.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
13. Moran C, Beare R, Phan TG, Bruce DG, Callisaya ML, Srikanth V, et al. Type 2 diabetes mellitus and biomarkers of neurodegeneration. Neurology. 2015;85(13):1123–30.
14. Li W, Risacher SL, Gao S, Boehm SL, Elmendorf JS, Saykin AJ. Type 2 diabetes mellitus and cerebrospinal fluid Alzheimer’s disease biomarker amyloid β1-42 in Alzheimer’s Disease Neuroimaging Initiative participants. Alzheimers Dement. 2018;10:94–8.
15. Qiu C, Sigurdsson S, Zhang Q, Jonsdottir MK, Kjartansson O, Eiriksdottir G, et al. Diabetes, markers of brain pathology and cognitive function: the Age, Gene/Environment Susceptibility-Reykjavik Study. Ann Neurol. 2014;75(1):138–46.
16. Wang R, Fratiglioni L, Kalpouzos G, Lövdén M, Laukka EJ, Bronge L, et al. Mixed brain lesions mediate the association between cardiovascular risk burden and cognitive decline in old age: A population-based study. Alzheimers Dement. 2017;13(3):247–56.
17. Moran C, Beare R, Wang W, Callisaya M, Srikanth V, Alzheimer’s Disease Neuroimaging Initiative (ADNI). Type 2 diabetes mellitus, brain atrophy, and cognitive decline. Neurology. 2019;92(8):e823–30.
18. Dufouil C, Dubois B, Vellas B, Pasquier F, Blanc F, Hugon J, et al. Cognitive and imaging markers in non-demented subjects attending a memory clinic: study design and baseline findings of the MEMENTO cohort. Alzheimers Res Ther. 2017;9(1):67.
19. Morris JC. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology. 1993;43(11):2412–4.
20. Hugonot-Diner L. MMS version consensuelle GRECO. In: La consultation en gériatrie. Paris: Masson; 2001. p. 13–20.
21. Grober E, Buschke H, Crystal H, Bang S, Dresner R. Screening for dementia by memory testing. Neurology. 1988;38(6):900–3.
22. Thurstone LL. Psychophysical analysis. By L. L. Thurstone, 1927. Am J Psychol. 1987;100(3–4):587–609.
23. Benton AL, Varney NR, Hamsher KD. Visuospatial judgment. A clinical test. Arch Neurol. 1978;35(6):364–7.
24. Tombaugh TN. Trail Making Test A and B: normative data stratified by age and education. Arch Clin Neuropsychol. 2004;19(2):203–14.
25. Operto G, Chupin M, Batrancourt B, Habert M-O, Colliot O, Benali H, et al. CATI: A Large Distributed Infrastructure for the Neuroimaging of Cohorts. Neuroinformatics. 2016;14(3):253–64.
26. Ashburner J, Friston KJ. Unified segmentation. NeuroImage. 2005;26(3):839–51.
27. Chupin M, Hammers A, Liu RSN, Colliot O, Burdett J, Bardinet E, et al. Automatic segmentation of the hippocampus and the amygdala driven by hybrid constraints: method and validation. NeuroImage. 2009;46(3):749–61.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
28. Desikan RS, Ségonne F, Fischl B, Quinn BT, Dickerson BC, Blacker D, et al. An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. NeuroImage. 2006;31(3):968–80.
29. Samaille T, Fillon L, Cuingnet R, Jouvent E, Chabriat H, Dormont D, et al. Contrast-based fully automatic segmentation of white matter hyperintensities: method and validation. PloS One. 2012;7(11):e48953.
30. Fazekas F, Barkhof F, Wahlund LO, Pantoni L, Erkinjuntti T, Scheltens P, et al. CT and MRI rating of white matter lesions. Cerebrovasc Dis. 2002;13 Suppl 2:31–6.
31. Habert M-O, Marie S, Bertin H, Reynal M, Martini J-B, Diallo M, et al. Optimization of brain PET imaging for a multicentre trial: the French CATI experience. EJNMMI Phys. 2016;3(1):6.
32. Buchert R, Wilke F, Chakrabarti B, Martin B, Brenner W, Mester J, et al. Adjusted scaling of FDG positron emission tomography images for statistical evaluation in patients with suspected Alzheimer’s disease. J Neuroimaging. 2005;15(4):348–55.
33. Toussaint P-J, Perlbarg V, Bellec P, Desarnaud S, Lacomblez L, Doyon J, et al. Resting state FDG-PET functional connectivity as an early biomarker of Alzheimer’s disease using conjoint univariate and independent component analyses. NeuroImage. 2012;63(2):936–46.
34. Habert M-O, Bertin H, Labit M, Diallo M, Marie S, Martineau K, et al. Evaluation of amyloid status in a cohort of elderly individuals with memory complaints: validation of the method of quantification and determination of positivity thresholds. Ann Nucl Med. 2018;32(2):75–86.
35. de Medeiros K, Robert P, Gauthier S, Stella F, Politis A, Leoutsakos J, et al. The Neuropsychiatric Inventory-Clinician rating scale (NPI-C): reliability and validity of a revised assessment of neuropsychiatric symptoms in dementia. Int Psychogeriatr. 2010;22(6):984–94.
36. Sanfilipo MP, Benedict RHB, Zivadinov R, Bakshi R. Correction for intracranial volume in analysis of whole brain atrophy in multiple sclerosis: the proportion vs. residual method. NeuroImage. 2004;22(4):1732–43.
37. Bollen KA. Structural Equations with Latent Variables. John Wiley & Sons; 2014. 474 p.
38. Rosseel Y. lavaan: An R Package for Structural Equation Modeling. J Stat Softw. 2012;48(1):1–36.
39. Schermelleh-Engel K, Moosbrugger H, Müller H. Evaluating the Fit of Structural Equation Models: Tests of Significance and Descriptive Goodness-of-Fit Measures. Methods of Psychological Research Online. 2003;8(2):23–74.
40. Jack CR, Bennett DA, Blennow K, Carrillo MC, Dunn B, Haeberlein SB, et al. NIA-AA Research Framework: Toward a biological definition of Alzheimer’s disease. Alzheimers Dement. 2018;14(4):535–62.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
41. VanderWeele TJ. Mediation Analysis: A Practitioner's Guide. Annu Rev Public Health. 2016;37:17-32.
42. R Core Team. R: A language and environment for statistical computing. [Internet]. R Foundation for Statistical Computing, Vienna, Austria; 2018. Available from: https://www.R-project.org/
43. Espeland MA, Bryan RN, Goveas JS, Robinson JG, Siddiqui MS, Liu S, et al. Influence of type 2 diabetes on brain volumes and changes in brain volumes: results from the Women’s Health Initiative Magnetic Resonance Imaging studies. Diabetes Care. 2013;36(1):90–7.
44. Moran C, Beare R, Phan T, Starkstein S, Bruce D, Romina M, et al. Neuroimaging and its Relevance to Understanding Pathways Linking Diabetes and Cognitive Dysfunction. J Alzheimers Dis. 2017;59(2):405–19.
45. Thambisetty M, Jeffrey Metter E, Yang A, Dolan H, Marano C, Zonderman AB, et al. Glucose intolerance, insulin resistance, and pathological features of Alzheimer disease in the Baltimore Longitudinal Study of Aging. JAMA Neurol. 2013;70(9):1167–72.
46. Benedict C, Grillo CA. Insulin Resistance as a Therapeutic Target in the Treatment of Alzheimer’s Disease: A State-of-the-Art Review. Front Neurosci. 2018;12:215.
Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.
DOI 10.1212/WNL.0000000000012440 published online July 1, 2021Neurology
Eric Frison, Cecile Proust-Lima, Jean-Francois Mangin, et al. Diabetes Mellitus and Cognition: A Pathway Analysis in the MEMENTO Cohort
This information is current as of July 1, 2021
ServicesUpdated Information &
ullhttp://n.neurology.org/content/early/2021/07/01/WNL.0000000000012440.fincluding high resolution figures, can be found at:
Subspecialty Collections
http://n.neurology.org/cgi/collection/petPET
http://n.neurology.org/cgi/collection/mriMRI
http://n.neurology.org/cgi/collection/cohort_studiesCohort studies
http://n.neurology.org/cgi/collection/cognitive_agingCognitive aging
http://n.neurology.org/cgi/collection/alzheimers_diseaseAlzheimer's diseasecollection(s): This article, along with others on similar topics, appears in the following
Permissions & Licensing
http://www.neurology.org/about/about_the_journal#permissionsentirety can be found online at:Information about reproducing this article in parts (figures,tables) or in its
Reprints
http://n.neurology.org/subscribers/advertiseInformation about ordering reprints can be found online:
0028-3878. Online ISSN: 1526-632X.Kluwer Health, Inc. on behalf of the American Academy of Neurology.. All rights reserved. Print ISSN:is now a weekly with 48 issues per year. Copyright Copyright © 2021 The Author(s). Published by Wolters
® is the official journal of the American Academy of Neurology. Published continuously since 1951, itNeurology